AI-201 · Question #309
An Agentforce agent must retrieve data from Salesforce and an external system. What is the recommended approach?
The correct answer is B. Use Apex or Flow with external integrations. Agentforce agents invoke actions (skills) to interact with data. For external system data, the recommended pattern is to create an action backed by an Apex class (using HTTP callouts) or a Flow that calls an external service via an HTTP Action or named credential. This keeps…
Question
An Agentforce agent must retrieve data from Salesforce and an external system. What is the recommended approach?
Options
- ATrain the LLM on external data
- BUse Apex or Flow with external integrations
- CCopy external data into prompts manually
- DDisable grounding
How the community answered
(24 responses)- A13% (3)
- B75% (18)
- C4% (1)
- D8% (2)
Explanation
Agentforce agents invoke actions (skills) to interact with data. For external system data, the recommended pattern is to create an action backed by an Apex class (using HTTP callouts) or a Flow that calls an external service via an HTTP Action or named credential. This keeps the integration logic inside Salesforce's governance model, respects security boundaries, and is maintainable. Training the LLM on external data (A) is not how grounding works and would be ineffective for real-time data. Copying external data into prompts manually (C) is impractical, insecure, and violates context window limits. Disabling grounding (D) would make the agent less accurate, not more capable.
Topics
Community Discussion
No community discussion yet for this question.